★ THE BLOG ★ Ramblings on WiFi & stuff.

Bad Design at Your Request

Estimated reading time: 6 minutes, 44 seconds. Contains 1348 words

 

What does one do when presented with a highly questionable request from a customer? Nothing immoral here, just when a potential customer is asking you to do something you know won’t work. I had this exact scenario happen this Summer with a resort that wanted to do a wi-fi refresh.

Originally, the customer just wanted to do a rip-n-replace - swap out their existing 7-year old, 2.4GHz APs - with new APs. After some discussion we convinced them that a simple swap out was not the best solution. We agreed to do a predictive survey using data collected from on-site data collection.

One of the caveats was that the APs could not be in the rooms. For aesthetic reasons, and others, management wanted no APs in rooms. We knew this would not be an ideal solution, and we let this be known on several occasions. After explaining our reasoning, IT was in agreement with us on this matter. However, the management was not convinced and decided to take a chance on the hallway “design”.

We did our best in collecting data on-site and used that data in our predictive surveys. In the end the initial deployment was a hallway placement. We adjusted some AP locations and added some, but I knew this solution would not yield the desired results. Directional antennas brought the budget to more than they wanted, so those were not an option. They also wanted this in on a tight time-line. I had made my concerns know at multiple occasions, but there was no budging from management.

So, I could choose not to do this project and walk away, or sell a solution I was not confidant in. Well, I have a business to run, bills to pay, and employees to… employ. I chose to “design” a solution within the constraints - both AP placement and budgetary. With this in mind I drafted the cover letter below for the predictive survey I presented to the customer:


Thank for this opportunity to allow us to present you with a wireless access solution for your wonderful property. Before you proceed to the information in the document please indulge us and read this overview in its entirety. It will clarify the purpose and scope of this document. 
 
This report is a “Predictive Survey”. The term “predictive” is used deliberately to denote the fact that the process used in the creation of this report is a best guess and will most likely not be 100% accurate. In a structurally complicated deployment such as yours we can probably assume a 75-80% accuracy rate. 
Predictive surveys are a very useful tool for the Wireless LAN Professional when a full, cost-prohibitive survey is unavailable. There are two methods to perform a predictive survey: 
 
OPTION 1. Using only the floor plans and a questionnaire we can use the survey software to automatically place the APs and then manually adjust to our specifications. We can alternatively manual add the APs to specification. This type of survey is best for traditional, modern open-floor plan office environment where the loss and performance characteristics are well known. This is also the most cost-affective (up font) solution and very commonly used. In the end you get the best data out when you put the best data in so this option should be used sparingly and mainly for budgetary purposes.
 
OPTION 2. Perform a physical site survey to become aware of the building materials in use, current AP locations and limitations, physical build of the rooms and furniture materials and locations, etc. Also, using the same APs and antennas that are being proposed take as many readings as possible to determine the true signal loss at varying distances from the AP. We also are able to look at various types of RF interference that may be present, and possibly mitigate that interference before the deployment begins. This allows us to use that data to better build our predictive model. This is less cost effective than option 1, but more cost-effective than a full site survey, and allows the model to go in with the best possible data that the circumstances allow. 
 
Option 2 is what we have done. We spent time on site taking readings, and noting material types as we could. Obviously, in an environment as busy as yours it was not possible to have access to every location so we did the best we could taking readings in multiple rooms and room types and through the various material types at your location. This is not perfect, nor definitive, but it will at least give us valuable information to use when making our predictive model. 
 
We would also like to take this moment to also state that locating all the APs in the hallways is the least effective model in a multi-room, multi-tenant environment such as yours. The best-case scenario here is allowing the APs to be located with-in the rooms. This allows us to use the structure of the building itself to allow for separation between the APs and help mitigate interference as well as getting the RF signal closer to the clients. We understand that this is not always possible for a variety of reasons, but we felt the need to make you - the customer - aware of the limitations to which this design has been restricted.
 
In conclusion, we ask that you look at the following report with the information in this overview in mind and with the understanding that after deployment we highly recommend that we (or a 3rd party) perform a validation survey to confirm where the predictive model falls short. Upon the results of this verification there may be several things that need to be done. It may be adding, relocating, or even removing some APs, or we may simply need to disable certain AP radios to reduce co-channel interference. Either way the design is not complete in our view if it has not been validated after implementation.
 
With this in mind please proceed to review the report. Thank you. 
 
Eddie Forero, Principal CommunicaONE Inc. 

The report essentially showed what we had been saying - that APs in the hallways would not provide the in-room coverage they desired. We also provided an alternative design with APs in room. In the end the management went ahead with the hallway solution despite ITs misgivings.

The end result was not much better than what they had. I fully expected to bear the wrath of the customer. I was not happy that I installed a solution I didn’t believe in. And I was not expecting what happened next.

The Director of IT flew out the the head office to present the results of our post-installation validation survey. He showed that hallway APs were providing great “coverage” in the hallways, but not in the guest rooms. He explained how we had predicted that this design would not give them the results they were after and the gamble did not pay off.

Because we had been very clear about our concerns, and because we had predicted, then validated those concerns, the management decided to foot the bill for a complete in-room redesign (even using less expensive APs). And not only that, but also light up another property next door!

Maybe we should have walked away. But, instead, I stated clearly why the solution would not work and made sure they were aware of a drawbacks. I’m don’t know if I would do this again, but I will definitely make even more of an effort in the future to have the customer deploy the right solution the first time around. It’s more cost effective and less stressful.

I don’t know if this is helpful to anyone, but I figured I’m not the only one who has had projects where your hands were tied. The moral of this story is - stand your ground. In this case it worked out because the customer realized the error and stepped up to do it right. But, make sure you fully layout the issues. Be respectful of the customer, but respect your skills and knowledge as well.